The CLIO collaboration has started in 1986 in order to build an infrared laser at LURE at Orsay. It includes the design and construction of a new FEL dedicated rf linear accelerator. The goal is to make a broadly tunable laser (at least in the range 2–20 μm), of high peak power (MW range), and “medium” average power (a few tens of watts). It will be utilized as a “user facility” as well as for FEL fundamental studies. The linac uses a gridded disperser cathode gun, a 500 MHz prebuncher, a fundamental buncher and one accelerating section powered by a 3 GHz klystron (with 12 μs long pulses). The final energy will be adjustable between 30 and 75 MeV. The desired characteristics are: peak current >50 A, total energy spread ≈1%, normalized emittance <150 πmm mrad. The numerical simulations made with PARMELA indicate that these performances will be reached with an emittance better by a factor of 2–4. The accelerator is followed by an achromatic and nearly isochronous 60° bend. It will allow to select the particles within a given energy spread, before the undulator, and to analyse, with respect to time, their energy distribution after the undulator. The optical cavity is 4.8 m long (32 ns roundtrip time), compatible with the laser risetime (typically 100 passes) and the klystron pulse length of 12 μs. The undulator is made of two independently adjustable identical parts, 0.96 m long (24 × 4 cm) each; the second one can be tapered. This will allow to optimize both the optical gain and energy extraction at different wavelengths. The laser is scheduled to operate by the end of 1990.